The Indeterminacy of Arms Length

could account for profitability differences, such as high start­up costs in the US market. Attempting to defuse political pressures, the Chancellor of the Exchequer John Major wrote to the Treasury Secretary Nicholas Brady proposing a joint multilateral study of tax avoidance by multinationals, an offer which Brady gratefully accepted, along with a similar German proposal. 1 What was most striking was that neither the efforts by the US over three decades to develop an effective transfer price regime for US foreign direct investments, nor the work in international bodies especially the OECD since 1975, were regarded as providing an adequate answer to US political concerns about the adequacy of taxation of inward direct investment.

3. The Indeterminacy of Arms Length

The administration of transfer pricing rules under the Arms Length principle has faced two basic difficulties. First, it entails a focus on specific transactions; and second, it relies on finding comparable transactions between unrelated parties. Underlying these is the basic problem that Arms Length depends on treating related companies as if they were unrelated. The essential advantages of a corporate group are that fixed costs can be jointly shared, and that a successfully integrated firm generates synergy profits, or additional returns attributable to the organization as a whole rather than any particular unit. The Arms Length approach based on transactional analysis entails attempting to dissect this unity. 3.a Arms Length Price or Arms Length Profit The Arms Length principle essentially requires adjustments to be made to the pricing of specific transactions rather than performing a global profit allocation. Nevertheless, both tax administrators and courts have used the profit split produced by price adjustment as an important factor either in validating price adjustments, or even to decide what the adjustment should be. Some have justified this by arguing that arms length merely requires affiliates to be treated as separate entities, and their assessment to begin from separate accounts, and that the approach is not vitiated by using profit split as a check Hunter, in Competent Authorities 1986, p.594. i Identification of Specific Transactions The US regulations of 1968 were the first to establish a detailed approach to the identification of specific intra­firm transactions in which one affiliate is the supplier of something to another. However, a supplier­receiver relationship may be hard to 1 Letter of 29 August, printed in Tax Notes of 10 September 1990. define among related companies sharing factor inputs. It appears most clearly in the sale of tangibles, since there is a clear transfer of a physical item. The supply of tangibles for use leasing may also appear to be a clear bilateral transaction, although there may frequently be shared use. A loan may also appear to be a clearly bilateral relationship, but the common case of a parent company guarantee for a subsidiarys borrowing is not easy to classify. Most problems are caused by services and intangibles, which commonly consist of activities carried out for the joint benefit of all or many of the companies in a corporate group. Thus, the US s.482 rules provide detailed definitions to identify the `renderer ¶ of services, the `developer¶ of know­ how, and the beneficiary or recipient of each. Further, a relationship which appears to involve a straightforward bilateral relationship may require more detailed analysis to define the various transactions involved. Thus, one relationship may involve two or more transactions: a sale of tangibles such as the active ingredient for a drug often involves a simultaneous transfer of intangibles, the rights to the patents embodied in it: therefore, `unbundling ¶ the different transactional elements may be important. Equally, a transaction may be reciprocal: a patent licence will commonly include a grantback clause for development patents, which must be dealt with by provisions for set­offs. ii Profit­Split: Criterion or Check? As the limitations of the transactional approach in the US Regulations have become more apparent, the use of profit­split has gained ground. Initially, there was uncertainty as to its validity: in Du Pont v. US, the trial judge stated critically that `the regulation approach seems to rule out net profit as a relevant consideration in the determination of an arms length price, this despite Congress encouragement to the contrary ¶ 1 However, the appeals court in the same case approved `consideration of net profits in appraising the realism of prices charged ¶Du Pont v. US 1979, p. 456. This seemed to make sense of the prior cases: in Lufkin Foundry 1972 the use of profit split by the Tax Court had been overturned on appeal, on the ground that there had been no attempt first to find an arms length price; while in Pittsburgh Plate Glass Industries Inc. 1970 a profit split analysis was used in support of the adjustment produced by a comparable. Finally, the courts went a stage further in Eli Lilly v Commissioner 1988, in accepting an allocation based on profit split, after attempts to fix arms length prices had failed due to non­existence of comparables. 2 A study of all the s.482 cases in the courts between 1962 and 1980 ­ 22 cases ± also 1 Judge Willi, in Du Pont v. US, cited in US Treasury 1988 p. 34. 2 In the meantime, Congress had enacted a new Tax Code provision, in response to this case, specifying that corporations with affiliates in US possessions i.e. Puerto Rico earning income from intangibles must divide the income, either under a cost sharing method, or on the basis of a 50­50 profit split: Tax Equity and Fiscal Responsibility Act 1982 s.213a, adding section 936h to the Tax Code; see below. confirmed the importance of profit­split Donnelly 1986. The position thus reached in the US was that profit split is important as a check on the reasonableness of adjustments to transaction prices; and it could also be used in its own right where adjustment based on arms length is not possible. In practice, in most countries, it is more likely to be the taxpayer that relies on the arms length price criterion, by pointing to a market price which should be accepted as comparable; while the main point of reference for the tax authorities will be the profit which the price produces. Although this means primarily a comparison with the profits of other similar undertakings, inevitably there will also be some consideration of the internal profit­split within the TNC. Thus, the French law of 1933 see s. 1.b above already provided that, in the absence of specific data to establish an arms length price, the assessment may be done by comparison with the profits of similar undertakings normally managed. Similarly, the Japanese law of 1986 states that if a company does not provide the information necessary to calculate the arms length price, the local tax office may calculate a price based on the gross margin ratio on sales or some other appropriate ratio of a similar company in the same line of business. A senior Japanese official has also indicated that profit split is likely to be used by administrators as a check on the `reasonableness ¶ of a price adjustment Competent Authorities 1986, p.591. Equally, the German law of 1972 s.13, as amended in 1976 allows, in the absence of other appropriate criteria, an assessment based on the return on invested capital, or margin on turnover, which could be expected under normal circumstances. The 1983 German regulations also specify that it is legitimate to compare the business results of the tax­payer, its related companies, and other similar companies, either as a check on or an alternative to other methods of establishing arms length prices. 1 Although the UK statutory provisions only permit adjustments to the pricing of specific transactions, in practice the profit split produced is a major factor, and a senior official has accepted that a formula approach `may well have a role to play in a final settlement based on economic reality ¶Hunter, in Competent Authorities 1986, p.594. An English court has held that the arms length rule in a treaty provision based on Article 72 of the OECD model does not require `the substitution for every transaction between the branch and the main enterprise of the transaction that would have been entered into if the branch and the main enterprise had been carrying on wholly separate businesses ¶ but `permits a method of ascertaining the profit to be attributed to the branch which is one which might have been agreed between the 1 Section 2.4.5 and 2.4.6, and Höppner 1983 p.222. Cf. also Saunders 1989 p.253, who states that the German criteria `basically require the German tax administration to consider that every company must make an adequate return on capital employed or a gross profit which is compatible with that of its compe WLWRUV¶ branch and the main enterprise if the branch had been an independent enterprise ¶ Vinelott, J. in Sun Life Assurance Co. of Canada v. Pearson 1984, p. 507. Nevertheless, the basic statutory provisions ICTA 1988 s.770 are much more specific than the treaty wording in requiring adjustments to pricing of transactions, and might cause some difficulty for the Inland Revenue if a case were to come to court. However, profit­split is considered to be appropriate only as a guide and not as a criterion in itself. The difficulty is that, if it is not based on any rigorous evaluation of the consolidated accounts of the group of related companies, it is no more than a rough check on the profitability of the local subsidiary in relation to the group and in comparison with other similar firms. There may frequently be good reasons for significant divergence. A TNC might limit its operations in particular markets to highly profitable lines, or may have superior technology, which should produce much higher profits than local firms. Conversely, a foreign­owned subsidiary may experience difficulty in establishing itself in a new market, or may for other good reasons register repeated losses, in contrast with the groups more profitable operations elsewhere. Furthermore, the use of profit comparisons has caused disagreements among national authorities. As we have seen, it has long been accepted, at least for a branch or other permanent establishment, that where separate accounts based on comparable market prices for specific transactions are not available, the profits could be calculated by comparison with similar independent firms, using percentage of turnover or other appropriate coefficients. Historically, this clearly meant comparison with similar enterprises located in the same country; but in the case of multinational banks, which frequently operate through foreign branches, Japan and the United States have taken the view that it is more appropriate today to use global coefficients or criteria see section 1 above, and OECD 1984­II paras 64­70. Finally, there is official agreement generally that the `separate enterprise ¶approach must be kept distinct from the unitary method, which if it has any formal validity is considered at most an exceptional and subsidiary method applicable only to permanent establishments. So long as this remains the official position, it is hard to find any formal justification for the use of `profit­split ¶in the sense of an evaluation of the profitability of a particular affiliate in relation to that of its corporate group as a whole. It is possible to argue that the separate enterprise criterion laid down in Article 9 of the model treaties does not require `arms length prices ¶for specific transactions, but an `arms length profit ¶for the affiliate, since it states in very general terms that where conditions between related enterprises differ from the separate enterprise standard, the profits which would otherwise have accrued may be restored and taxed accordingly. This justified the OECD Committee in treating the question of thin capitalization as a transfer pricing matter OECD 1987C­I discussed below, s.3.c.i. Significantly, however, the German authorities expressed a reservation to that reports reasoning, on the grounds that it was based on `the notion of an `arms length profit, rather than on the generally accepted notion of an `arms length price µ ibid., p.36 fn.2. This demonstrates a reluctance to move away from the apparent security of specific price adjustments towards profit comparisons. The argument that the relevant profit comparison must be with global industry averages, because local firms are not comparable, made by Japan and the US in relation to banking, may be made for other businesses. In such cases, the `arms length profit ¶ criterion might become a very loose approximation for a global profit allocation; and this the tax administrators certainly wish to avoid. Thus, national laws, administrative regulations, court decisions and official practice all emphasise that profit­split is an ancillary criterion, and the primary approach must be to discover market prices charged by independent parties dealing at arms length. 3.b Tangibles: The Search for Comparables The US regulations, in particular, explicitly specify the Comparable Uncontrolled Price as the primary method and touchstone for pricing each category of transaction. The search for comparables has been fiercest in relation to tangible sales, since they are the most clearly bilateral transactions; moreover, the primary alternative method, Resale Price Minus, also depends on discovering unrelated parties in comparable circumstances: `the vital prerequisite for applying the resale price method is the existence of substantially comparable uncontrolled resellers ¶ DuPont v. US 1979, p.450. i The US Experience More than twenty years of experience of administering the US regulations has shown that in a large proportion of cases where TNC pricing is examined no comparables can be found. This has been shown quantitatively by half a dozen studies of s.482 allocation cases. The Treasurys own report of 1973 revealed that of a total of 174 adjustments to the pricing of tangibles only 36, or 20, were on the basis of CUP see Tables 4 and 5. This proportion was confirmed by academic studies based on questionnaires sent to companies US Conference Board 1972, Burns 1980. Criticism of the Arms Length method reached a peak when the General Accounting Office GAO published a report to the Congress in 1981, which concluded: Because of the structure of the modern business world, IRS can seldom find an arms length price on which to base adjustments but must instead construct a price. As a result, corporate taxpayers cannot be certain how income on inter­corporate transactions that cross national borders will be adjusted and the enforcement process is difficult and time­consuming for both IRS and taxpayers. ... We recommend that the Secretary of the Treasury initiate a study to identify and evaluate the feasibility of Table 4 Methods Used in s.482 Price Adjustments for Tangibles Study Method used CUP Resale Cost Other Minus Plus 1. Treasury 1973 20 11 27 40 2. Conf. Bd. 1972 28 13 23 36 3. Burns 1980 24 14 30 32 4. GAO 1981 15 14 26 47 5. IRS 1984 41 7 7 45 6. IRS 1987 31 18 37 14 Source: US Treasury 1988, p. 22, derived from the following: 1. US Treasury 1973 reprinted in Murray 1981 308. 2. US Conference Board report no. 555 1972. 3. Burns 1980. 4. US General Accounting Office 1981, ch. 4. 5. US IRS 1984. 6. US Treasury 1988, Appendix A, p. 8. Table 5 Official Studies of s .482 Price Adjustments Treas. 1973 GAO 1981 IRS 1984 No. of Corps. Or Case Files 519 823 Files with potential adjs. 871 Corps. For which adjs. made 200 Transactions for which adjs. considered 1706 3080 Adjs. Made 886 403 2306 Adjs. Agreed Value of adjs. In m 520 662 277 1335 4377 TANGIBLES Value in m 313 124 2632 Potential Adjs. 591 589 Adjs. Made 174 34 339 of which agreed 91 127 METHODS USED No. No. No. CUP 36 20 5 15 139 41 RM 19 11 4 14 24 7 CP 48 27 9 26 22 7 Other 71 40 16 47 154 45 Sources: as for Table 4, lines 1, 4 and 5. The 1973 study was based on returns for which audit was completed in 1968 and 1969; the 1981 study cases closed in 1978 and 1979; and the 1984 study those closed in 1980 and 1981. ways to allocate income under s.482, including formula apportionment, which would lessen the present uncertainty and administrative burden created by the existing regulations ¶US GAO 1981 p.54. The IRS and Treasury strongly countered the conclusions of the GAO report, and the figures on which they were based. There was certainly some basis for attacking the validity of the GAOs extraordinarily low figure of 15 for adjustments based on CUP, since the total of 34 adjustments to the pricing of tangibles was not a large statistical sample. But what was surprising was that as few as 34 adjustments had been made to tangibles pricing, in a programme of audit including all TNCs with assets over 250m, in which over 500 International Examiners reports were completed over a two year period. In contrast, many more adjustments were made to the pricing of loans, services and rents ­ 274 of the 403 adjustments; but they were almost all 240 or 87 made using safe haven rules. However, these accounted for only 60.4m by value, compared to the 277m produced by only 34 adjustments to tangibles pricing. The clear implication is that verification of tangibles pricing, since safe haven rules could not be used, consumed more time and expertise; but equally, it could produce more significant adjustments. The IRS claim that pricing adjustments for tangibles were based on CUP in over 40 and perhaps as many as 50 of cases was subsequently supported by its own comprehensive study published in 1984, based on cases closed in the two years following those covered by the GAO study. Nevertheless, even this still showed that adjustments were more often made by `other ¶methods than by using the CUP. The most striking factor was the sharp increase in the number of adjustments made: although the number of case files only increased from 519 to 823 1.6 times, the overall number of adjustments increased nearly six­fold, with a total value over 15 times as large; while for tangibles, adjustments increased ten times in number and over twenty times by value Table 5. Clearly, the IRS had taken to heart the admonitions to improve enforcement of s.482, including the GAOs recommendation to use economists in all major cases not involving safe harbours Treasury 1988, p.25. In fact, the Treasury changed tack, and the 1988 White Paper no longer attempted to defend previous practice, but proposed a new approach to pricing, the Basic Arms Length Return Method see below. The statistical evidence is a significant indication that there are serious difficulties in establishing arms length prices on the basis of comparables. The evaluation of detailed qualitative factors, as well as the effect of quantity on price, make it very difficult if not impossible to establish with certainty the comparability of similar items sold by third parties, and even of the same item sold by the same firm to third parties. Above all, few tax authorities have the time or expertise for this process. Even the USA has had difficulty in allocating resources commensurate with the growing scale of the problem: the total number of international examiners grew from 150 in 1977 to 505 in 1988, but this was merely keeping pace with the growth of US companies abroad, and the even more rapid expansion of foreign­owned businesses in the US Woodard 1988. ii The Administrative Burden of Scrutiny Whether for political reasons or merely through lack of resources, other countries have put much less effort, even comparatively, into the scrutiny of transfer prices. In the UK, the transfer pricing unit of the Board of Inland Revenue 1 was reported in 1984 to have 7 staff only one of whom was a qualified accountant. Nevertheless, this unit was said to have been responsible for transfer price adjustments to profits `of the order of £200m ¶ in the decade 1974­1984; while in the single year 1987­88 settlements of transfer pricing cases gave an immediate tax yield of £71m, and Controlled Foreign Company cases £12.5m. 2 In Germany the federal system means that tax assessment and collection is done by the Länder, of which only Bavaria has inspectors with specialised training, although coordination is ensured by a Federal Finance Office with some 70 auditors of whom about 10 specialise in international matters. 3 However, due to differences in the administration of tax enforcement, direct international comparisons of the resources and effectiveness of departments responsible for transfer price adjustments are hard to make. Developing countries in particular have recognised that it is crucial to establish some monitoring especially of import prices paid, not only on related company purchases, and not only for tax reasons but even more importantly to safeguard valuable foreign exchange. The establishment of a government monitoring unit, although expensive, is likely to be highly cost­effective, as was shown by countries such as Greece in the late 1970s Ganiatsos 1981, UN 1978A; however, this type of unit is likely to be vulnerable to political pressures. especially in a period of liberalization and encouragement of foreign investors. 1 At that time, Technical Division 2B; subsequently a single International Division has been created, including both Policy and Technical work, and the Transfer Pricing unit is now International 5B: see UK Board of Inland Revenue 1988. 2 The figures for staffing in 1984 and yield 1974­84 are from a ministerial answer to a parliamentary question, Hansard 1983­4 vol.52 p.222; that for 1987­88 is given in the Management Plan of the International Division, published in UK Board of Inland Revenue 1988 vol.3. The latter document remarks: `With regard to transfer pricing and Controlled Foreign Company casework the investigations handled on the Section are generally into major multinationals and are complex and specialised, often taking up to 5 years to complete with an uncertain outcome. It is also possible that some Transfer Pricing cases may finally reach the Special Commissioners over the next three years, dependent on the resources of the Solicitors Office to offer advice etc. The result is that yield fluctuates year by year, FDQQRWEHPHDQLQJIXOO\IRUHFDVWDQGEHDUVQRUHODWLRQVKLSWRWKHVWDIILQJUHVRXUFHVSXWLQWRWKHP¶ ibid 9.9. It should be borne in mind that, some tax districts, notably those covering the City of London, and the special Oil Taxation Office, deal directly with many cases, and have considerable expertise of their own, so that the central unit is mainly concerned with training, advice and international liaison, and is directly concerned only with major or difficult cases. 3 Information from German Ministry of Finance, April 1987. In the1980s, as the problems of sovereign debt and shortage of foreign exchange have come more to the fore, many states have preferred to employ private firms to provide pre­shipment inspection services. This approach has been subject to political and other criticisms, but in view especially of the currency and capital flight problems of developing countries, it has become well established. Such a system can greatly facilitate the practical task of tax authorities. The need for confidentiality need not prevent information gathered for the purposes of one branch of government, such as customs or the Central Bank, being used by another; and although there must be some limitations on tax authorities divulging information from tax examinations, exchange of information between government branches, subject to safeguards, is essential for efficient administration. For example, import prices declared to Customs may differ from those returned to Tax authorities, since the incentive to inflate invoices to reduce taxable profits or transfer funds abroad may conflict with the desire to reduce import duties. 1 3.c Safe Harbours and Intra­firm Financial Flows. Some of the difficulties involved in establishing comparables may be avoided by the use of safe harbours. Safe harbours have two major advantages: they are easy to administer; and if they can be internationally agreed, they can ensure that an arms length adjustment by one state will be automatically compensated by a corresponding adjustment by the other to prevent economic double taxation. However, a safe harbour is only possible for standardised transactions taking place in competitive markets. In such conditions, it may be possible to define a reasonably precise rule; but there is still the issue of whether such a rule should be inflexible, or merely create a rebuttable presumption. Fairness may require that the tax­payer be allowed to demonstrate that a price charged complies with the arms length criterion even though it is outside the safe harbour. But from the point of view of the government this `would serve only to reduce tax liability ¶ since taxpayers would normally choose to rebut the presumption if it would be advantageous US Treasury 1988, p.73, although administrative costs must also be borne in mind. To allow both the taxpayer and the revenue to rebut the presumption would create uncertainty and defeat the purpose of the safe harbour. Although some administrations do use a `rule­of­thumb ¶ usually of an informal type, in some situations, this can hardly qualify as a safe harbour. Loans are relatively standardised transactions taking place in highly competitive markets, and would therefore seem suitable for a safe harbour approach. For international lending, however, this means deciding which market should provide the 1 In the UK, the Keith Committee on Revenue enforcement powers recommended increased information exchange between the Inland Revenue and other government branches, especially the Customs and Excise, which is responsible not only for import duties but also for excise taxes including VAT. reference point. It has proved difficult to settle this in terms of principle, largely because there is an inevitable divergence of perspective between the source country, which is concerned to ensure that high interest rates on inter­affiliate loans are not used to reduce the taxable business profits of the local subsidiary, and the parents country of residence, which will seek a full return on loans of capital. The UN Group of Experts went furthest, in specifying that the focus should be on the creditor, and therefore interest should be based on the creditors borrowing capacity and the interest rate either in its own country or on the relevant capital market Surrey 1978B, p.161. This reflected the viewpoint of developing countries as capital importers; they would be reluctant to allow deduction of the higher rates of interest that would be likely to prevail in their domestic markets, since foreign investors would benefit from lower rates in the international financial markets. The OECD report was however much more circumspect: it merely listed the criteria to be taken into account in determining what would be a comparable loan, and in relation to the appropriate interest rate it could only conclude that `no straightforward principle presents itself which would win general acceptance, and it would depend on the facts of the particular case ¶ OECD 1979, para.200. The US is the only major country which formally uses safe harbour interest rates which have only since 1988 been pegged to a realistic rate, published monthly. However, the home states of TNCs have had to accept that these firms borrow in the cheapest markets, usually off­shore, to finance their foreign subsidiaries. The home country revenue authorities have therefore shifted their attention from the rate of interest charged on inter­affiliate loans to ensuring that the capital structure of foreign subsidiaries realistically reflects the parents investment. Highly complex arrangements can be devised for financing international operations, and tax considerations are an important element. A key factor is that interest is treated, in most cases, as an expense which is deductible from business profits. Furthermore, tax treaties usually reduce withholding tax at source on interest to a low level, often zero. It is also possible to route such payments through a conduit company to a base holding company in a tax haven, thus ensuring that the profits are not taxed at all see Chapter 6 above. i Thin Capitalization Primarily for this reason, a TNC determining the capital structure of a subsidiary is likely to favour a high ratio of debt to equity; also because inter­affiliate interest payments may be more flexible and easier to repatriate than declarations of dividend. There may also be advantages in using forms of hybrid financing, such as convertible bonds or participating loans, which might be deductible for tax purposes while being treated as share capital from an accounting point of view. Under some circumstances a hybrid instrument might be treated as debt in the payors country so that the service payments are deductible interest, although they may be subject to withholding tax but as equity in the recipient state so that they are dividends and qualify for a foreign tax credit. However, the tax authorities may challenge such arrangements as being disguised equity contributions, or `thin capitalization ¶ This entails the rejection by the source country of the companys categorization of the nature of its investment, usually by disallowing deductibility of some or all of the interest, and perhaps also by recategorising some of the payments as dividends and taxing them accordingly. Although a state is free to take such action unilaterally under its own laws, it clearly raises international issues. A refusal to allow interest deductions for foreign­owned firms may be seen as an impediment to international investment. The model double tax treaties do not specifically deal with the matter, but several of the treaty provisions offer a possible basis for a state to object to a treaty­partner disallowing deduction of an interest payment, or reattributing such a payment as a dividend. Such a disagreement would fall to be resolved under the Mutual Agreement procedures of the treaty Article 25 of the Models, see Ch. 10.3 below. The thin capitalization issue has been addressed in general terms by the OECD Committee, which published a report favouring its treatment as a transfer pricing question under Article 9 of the Model treaties. 1 The argument for this is that, unless a states action is justifiable under article 9, it may be contrary to article 245, which requires non­discrimination between local and foreign­owned companies as regards deductibility of interest, royalties and other payments. However, if the thin capitalization rules apply equally to locally­owned companies, there is no discrimination and Article 245 does not apply. 2 In fact, there is nothing in the tax treaty models requiring a state to allow the deductibility of interest. If a state goes further than disallowing an interest deduction and reattributes a payment as a dividend, it might fall foul of Article 10, which limits source taxation of dividends to those defined as such in Article 103: income from shares ..., not being debt­claims, participating in profits, as well as income from other corporate rights which is subjected to the same tax treatment as income from shares by the law of the State of which the company making the distribution is resident. A minority of the OECD Committee considered that the exclusion of `debt­claims ¶ overrides the general inclusion of `other corporate rights ¶ and precludes reattribution of interest as dividends; but the majority considered that in appropriate cases reattribution was permitted; and it was generally agreed that the ambiguity should be 1 OECD 1987C­I, following up the brief discussion of the problem in the 1979 report, paras. 182­191. 2 Nevertheless, France has explicitly reserved the right to apply its domestic law on deductibility of interest notwithstanding Article 24: OECD 1987C­I para.66. cleared up in later versions of the treaty or its commentary. 1 If thin capitalization rules are treated as transfer price adjustments permitted under Article 9, rather than a valid exercise of the source countrys unilateral jurisdiction, then the implication is that they must comply with the Arms Length criterion. Nevertheless, some OECD member states apparently took the view that Article 9 could be read as being `illustrative ¶rather than `restrictive¶OECD 1987C­I, para 49 in this context. However, the general view of the Committee has been that Article 9 is definitive, in requiring transfer price adjustments to conform to Arms Length, and the general consensus here also was that `thin capitalization rules ought not normally to increase the taxable profits ... to any amount greater than the arms length profit ¶ Ibid. para.49. However, Germany entered a reservation to the reports use of the concept `arms length profit ¶ rather than the more specific principle of the arms length price, and pointed out that `the consensus regarding the actual application of the arms length principle is extremely vague and precarious ¶ibid. p.36.. Once again, however, the problem faced by the Committee has been whether to opt for a fixed rule, with the merit of certainty and predictability, or a flexible approach, which can respond to the particular circumstances of each case. A majority took the view that a fixed capitalization ratio would be inconsistent with the Arms Length requirement, unless it merely created a presumption which the tax­payer could rebut; and the report also urged that if a ratio is adopted, it should be fixed as high as possible, to minimise the number of taxpayers who would be obliged to take on the burden of disproving the presumption ibid. para. 79. Clearly, this type of safe harbour would have the disadvantages for the revenue identified by the US Treasury and mentioned above. On the other hand, the Committee recognised the wide discretionary scope of the case­by­case approach. Although much comparative information is available to tax authorities about company financing, a company can normally choose its capital structure from a wide range of options depending on the specific conditions. Furthermore, it may not be appropriate to ask what level of debt an independent third party, such as a bank, would have allowed a company, since such a party would not have available to it the amount of information which a parent would have about its own subsidiary ibid. para. 76. Thus, the Committee could come to no better conclusion than that the judgments of tax authorities and courts should be consistent and based on evidence of transactions between independent persons, applied in a reasonable manner para. 78. In practice however, it seems that revenue authorities do operate defined ratios as an informal guide, to avoid the uncertainty of a case­by­ 1 OECD 1987C­I. paras.56­60. The protocol to the France­US treaty signed in 1988 included a provision widening the definition of dividends to include any income treated as income from shares by the country of residence of the distributing company. case approach, as well as the time­consuming process of individual evaluations. 1 However, the validity of Revenue challenges to related­company thin capitalization based on analogies to third­party `arms length ¶situation has been directly challenged by the increased use by many companies of leveraged recapitalizations, which have made it very hard to define a `normal ¶capital structure Briffett 1990. Thus, the weapons available to revenue authorities are at best cumbersome, in relation to the complexity of financing devices open to companies produced by the fertile minds of the highly­paid accountancy and law firms. For example, the power to adjust related company financial transactions may be circumvented by channelling the loan through an unrelated company, back­to­back with a loan to it from a non­ resident affiliate. Further advantages can be gained by using dual­resident companies, whose accounts can be consolidated with those of related firms in two jurisdictions, thus allowing a `double dip ¶ or deduction of the same interest twice. 2 These are merely some of the simpler schemes devised by the international tax specialists. The complexity of financing arrangements makes it very difficult for the authorities to operate effectively with precise rules; but it is equally hard to apply broad anti­ avoidance provisions without accusations of arbitrariness. Once a particular form of financing has become established, a move against it by one state results in objections from the business lobbies that they are being unfairly treated and their competitiveness would suffer. Thus, the growth of debt­financed company acquisitions LBOs ­ leveraged buy­outs, many of them from abroad, led the US Congress to enact provisions in 1989 to limit deductibility of interest paid to related parties. These rules against `earnings stripping ¶provide for non­deductibility of `excess interest ¶paid to a related person if the recipient does not pay US tax on such income, and the payor corporation has a debt ratio over 1.5 to 1 and `excess interest expense ¶ interest payments above a certain proportion of income Mentz, Carlisle Nevas 1990. Business lobbies on behalf of overseas investors objected that this denial of interest deduction entailed a unilateral override of tax treaties see Chapter 11 below. On the other hand, the US measures at least have the merit of providing a specific statutory rule. A comprehensive solution would require an internationally coordinated approach. This could take the form of international agreement on more precise thin capitalization rules, although the increased complexity of company financing would make such rules arbitrary in operation, and probably discriminatory as between domestic and international transactions. A more radical approach would entail a joint move to end the deductibility of interest altogether, which might be coupled with an appropriate reduction in marginal corporate tax rates Bird 1988. While this has been 1 Thus, the UK Inland Revenue has been widely stated to operate an informal safe harbour debt­equity ratio of one­to­one, as well as a rule of thumb that pre­tax profits should be at least three times the interest expense: see Tomsett 1989 p.143. 2 This is being combated by provisions against dual residence. considered by individual countries for example, by the US Treasury in the course of the Reagan tax reforms a concerted decision would be difficult to achieve. ii Global Trading and Transfer Parking Special difficulties have increasingly arisen in dealing with the allocation of gains and losses of firms which are transnational financial intermediaries investment houses, both banks and stockbrokers operating in the key financial markets around the world. In principle, their activities can be dealt with according to the traditional separate enterprise approach, by attributing the profit from each transaction to the branch or subsidiary making the sale, unless it is considered to be a dependent agent for the parent or head office and therefore entitled only to a commission similar to the `merchanting profit ¶of a wholesaler. However, the growing complexity of multiple operations in different global markets, exploiting the possibilities of arbitrage and of market scope by trading assets continuously on an almost 24­hour per day basis, have undermined the effectiveness of the separate enterprise approach. Global trading offers considerable advantages of hedging and arbitrage exploiting even small price differences in different markets as well as access to a wider client base. An important element is regulatory arbitrage: exploiting differences in the regulatory treatment of transactions between markets. This includes not only or primarily tax treatment, but also various monetary and banking controls and prudential requirements, such as capital adequacy and liquidity rules, prohibitions of maturity mismatching and limits on foreign exchange exposure. A direct means of regulatory arbitrage is for global finance houses to exploit the possibilities of internal intra­firm trading by the `transfer parking ¶ of transactions such as loans and foreign exchange positions. The use of transfer parking became publicised by investigations into the activities of New Yorks Citibank, following allegations made by an employee in 1977, which were revealed in a court case for unfair dismissal, and were followed up by investigations by the SEC and a US Congressional Subcommittee. 1 Parking entails transferring a loan or foreign exchange position to a related branch or subsidiary, either by booking it directly to the related entity, or by undertaking an offsetting transaction, sometimes returning it also by a `round­trip ¶or `back­to­back¶procedure. In the highly competitive money markets, a margin of a fraction of one percent determines profit or loss, and structuring transactions to minimise the costs of regulatory compliance may become a necessity. Thus, Citibank in Paris registered a foreign exchange loss by instructing New York to direct the Nassau `branch ¶merely a separate set of books in New York to buy 6m at FFr 4.7275, and then to resell the 1 The material is usefully summarised by Sarah Bartlett 1981, and in Richard Dale 1984, Appendix 2, `Citibanks `Rinky LQNHDOV DVH6WXG\LQ5HJXODWRU\UELWUDJH¶ dollars to the New York and Brussels branches, for resale back to Paris, at FFr 4.7375. This reduced the taxable profits in Paris, and although the Nassau branch profits were taxable in the US, the banks excess foreign tax credits still made it worthwhile. In such cases, where there is evidence that the sale and repurchase were simultaneous, it might be possible to establish tax evasion. In the absence of such proof, the validity of this type of transaction could depend on whether the sales were within prevailing market ranges, and whether they were booked in a low­tax centre in anticipation of profit, or retrospectively once it was known that a profit and not a loss had been achieved Bartlett 1981. Citibanks transactions were also part of more complex arrangements whereby the Paris office could avoid French tax on foreign exchange profits, converting them into lower­taxed interest income, by swapping a deposit in a low interest currency for a high interest currency which was placed with a related branch, giving Paris a loss on foreign exchange from the forward discount on the high interest currency, but interest earnings from the interest rate spread between the two currencies Dale 1984, p.201. The potential tax advantages of linking interest rate and currency swaps have been summarised as follows: `Of course, with careful construction ... there are opportunities for linking swaps to high interest borrowings of a soft currency with resulting benefits, such as the creation of tax­free gains and a capital gains tax loss, all in a fully hedged situation ¶Selter, Godfrey Atkinson 1990. Tax authorities are not powerless against such manoeuvres: they can re­attribute a profit transferred by the re­booking of a transaction; or more radically, can re­ attribute all the profits made by an offshore office, on the grounds that it is effectively managed from the local branch. This step was indeed taken by the German authorities in relation to business booked to the Nassau unit of Citibank based in Frankfurt Dale 1984, p.199. However, they face major enforcement problems due to the sheer number of deals and the dispersed nature of financial markets. These markets are dominated by a network of major participants, so that the rates quoted in the multiplicity of bilateral deals result in a prevailing price range; but there is no single market and no universally recognised market rate. 1 As Edwards, the dismissed Citibank employee pointed out: `In order to get a full picture of a Citibank inter­bank transactions, it is necessary to examine its individual components at all branches involved. The chances of random sampling in several Citibank branches producing a single complete parking transaction are almost non­existent ¶cited in Bartlett 1981, p.110. Neither bank regulators nor tax authorities have the resources for continuous monitoring of such transactions on a world­wide scale. Ominously, Dale concludes that such activities `far from being confined to Citibank, are endemic to multinational banking ¶Dale 1984, p.204. 1 Report of Citibanks Audit Committee, prepared by the law firm Shearman and Sterling and accountants Peat, Marwick and Mitchell, cited in Bartlett 1981, p.111. Not surprisingly, it has been difficult for the national authorities to develop their own approaches to taxation of the profits from global financial trading, let alone establish agreed principles. The OECD Committee published one short study on taxation of foreign exchange gains and losses, which dealt only with gains or losses made incidentally in the course of a business which is not normally that of dealing in currencies, and even that report was described as a preliminary step in a rapidly developing field OECD 1988, para.5. On the treatment of currency swaps, the report was confined to a bare description para. 70, not mentioning intra­firm swaps, and stating that `The tax treatment of payments under instruments of these types is, in general, still being developed, and it is not yet, therefore, possible to comment on it very informatively ¶ibid. para 71. In the meantime, the possibility of wide divergences in tax treatment created significant uncertainty. In response to a request from the US IRS studies were published by specialists, both in private practice and in the government service Plambeck 1990; Ernst Young 1991, which disagreed as to the appropriateness of formula apportionment. However, whether an apportionment or a separate entity approach is adopted, it would require detailed agreement between the major taxing authorities on the characterization of transactions and the attribution of gains and losses between the elements of a global trading team based in different offices. The official position of national tax authorities, and the view expressed by a consultants report for the Institute of International Bankers, was suspicious of `arbitrary ¶ allocation by formula, and preferred profit­attribution by identification of the functions carried out and risks assumed by each party Ernst Young 1991; but the formula approach was argued to make more economic sense, by acknowledging that the profits of a globally­traded book of assets result from the synergy of the team as a whole Plambeck 1990. A formula approach would require prior agreement between the various parties, which critics argued would be hard to achieve, but was put forward in its favour as being administratively simpler; while administration of a pricing system based on the separate entity approach would be facilitated by an advanced ruling agreement. iii Proportionate Allocation of Financing Costs Transfer parking is primarily a problem in relation to banks and other financial enterprises. However, many manufacturing TNCs are also major financial institutions in their own right. They are also of necessity active in foreign exchange markets, and the definition and allocation of their tax base will be significantly affected by the approach under national tax rules to currency conversion and foreign exchange gains and losses OECD 1988. This is therefore another area where a concerted and coordinated approach by the major tax authorities is urgently required, to remove the inducement to firms to reduce their costs by international tax avoidance and arbitrage. A different approach to intra­firm finance would be to allow or require deduction based on the average cost of external borrowing for the firm as a whole. This position has been taken in the US rules which require allocation of interest expense of a single corporation between its domestic and foreign activities according to the ratio of domestic to foreign assets Internal Revenue Code s.864. This applies especially to banking, since transnational banks do much of their foreign business through branches. Moreover, the Tax Reform Act 1986 extended this to consolidated corporate groups, including multinational groups falling within a defined affiliation rule. This rule was fixed by the IRS in temporary regulations at 80 ownership. It was soon after revealed that the Ford Motor group had decided to reorganise its financial, insurance and leasing subsidiaries to be owned through a new holding company, 25 of the shares of which would be placed with institutional investors, so that the Ford ownership would fall below the 80 threshold. 1 However, the apportionment of interest expense has been criticised and rejected by a majority of the OECD Committee, which took the view that both a foreign branch and a subsidiary should be permitted to deduct the interest actually charged on inter­ affiliate or parent company loans, provided the rate is arms length OECD 1984­II, especially paras 58­62. Both Japan and the US took the view, opposed by the majority of OECD members, that Article 7 of the model treaty permits but does not require deductibility of the interest charged between bank branches and the head office. Japan only allows deduction of the actual inter­branch interest charged if the source of the funds can be traced and documented; but since this is not normally possible, the interest cost allowed is normally a reasonable estimate based on prevailing rates. The US approach takes the view that money is fungible ­ or at least money of the same currency. Foreign bank branches in the US are permitted to deduct interest based on the original source of funds only if that original price is the price of Eurodollar borrowing; otherwise, a `currency pools ¶method is used, which takes the average cost of borrowing for the corporation as a whole in that currency. This is defended as fairer and simpler, since `a banks receipts and payments of interest may be regarded as all flowing into and out of one common pool ¶ ibid. para.59. However, the Committee took the view that it was not simpler to operate, since the detailed calculation of the cost of borrowings would be complex. More importantly, it disregards the specific role of the particular branch, which might not be typical of the bank as a whole, and might result in different costs, and thus in higher or lower earnings. The broader problem is the anomaly entailed in apportioning or attributing costs on the basis of an average of the whole enterprise, while still purporting to tax the branch on a separate enterprise basis. In that respect, the question of the cost of money to a 1 Tax Notes 30 October 1989, p.531. banking firm is part of the larger problem of allocation of joint costs. 3.d Central Services: Joint Costs and Mutual Benefits A matter which has long been the subject of conflicting views between the home and host countries of TNCs is the allocation of charges in respect of central management and service costs. Indeed, the model treaties still include a provision, inherited from the League drafts, specifying that allowance must be made in the calculation of the business profits of a permanent establishment for `executive and general administrative expenses ¶ incurred on its behalf Article 73. However, the UN model explicitly limits the deduction to `actual expenses ¶in respect of management or other services or the use of patent or other rights; in contrast, the USA prefers to include in its treaties a specific reference to the deductibility in calculating branch profits of a `reasonable allocation ¶ of research and development as well as other expenses incurred for the purposes of the enterprise as a whole. For separately incorporated subsidiaries, there is no explicit international provision for dealing with such central service costs, but the starting point is the `separate enterprise ¶standard of Article 9. Nevertheless, it has been accepted that some of the general overhead costs of a multinational company group are incurred on behalf of the group as a whole, in respect of services provided to its members. The group may be structured so that such central services are provided either by the parent company, or by a regional centre for affiliates in that region, or by specific service centres such as distribution and marketing centres, a captive insurance company, or research laboratories. Such services may be administrative, for example planning and coordination, budgetary control, accounting and legal services, and computing; they may relate to staff matters such as recruitment and training; or they may relate to production in a broad sense, covering joint purchasing, distribution and marketing services including advertising, and research and development, including the administration and protection of intellectual property rights. From the point of view of TNCs, these represent costs which must be deductible somewhere against gross profits, and they consider that tax authorities should not interfere with any reasonable arrangement for sharing or allocating them. The tax authorities for their part fear that a TNC may allocate such costs `in order to obtain a tax advantage or to neutralise a tax disadvantage ¶OECD 1984­III para.7; and they are unwilling to accept unreservedly the principle that all such expenses must be deductible somewhere. 1 In principle, therefore, the attribution of such costs must be justified in terms of the 1 For example, costs connected with exempt income may not be deductible ­ OECD 1984­III para.18. arms length criterion. This raises two separate but interrelated questions: whether a charge can be justified, and what it should be. In principle, a charge can be justified only if the recipient can be shown to have derived a benefit. This `benefit test ¶is most easily satisfied if the charge is made directly for specific service transactions, on a fee­for­service basis. Preferably, such a fee should be verifiable as arms length by comparison with the cost of similar services from unrelated suppliers: for example, legal services could be billed at an hourly rate, which could be compared with the rate for comparable services from independent law firms. 1 In practice, however, direct fee­for­service charging is frequently difficult. The main reason is that central services often do not benefit one affiliate exclusively. This is obviously the case for services carried out for the joint benefit of all or several affiliates, for example an international advertising campaign. Indeed, it could be argued that even if on one occasion a specific service is carried out for a particular recipient, there may well be a subsequent benefit to others if knowledge or expertise is thereby acquired. Furthermore, the benefit may be indirect andor non­specific: it is common, for example, to pay a retainer to ensure the availability of some types of professional or specialised services, which may in the event not be required. Thus, it may be very difficult or impossible to evaluate the proportion of the benefit attributable to a particular affiliate; or to do so might require disproportionate administrative costs. It is therefore common for central service costs to be charged by an `indirect ¶method apportioning the costs. Most tax authorities have been obliged to accept this, although some only do so if direct charging is impossible, and an actual benefit can be shown. On the other hand, the US rules mandate US parent companies to make a cost­ apportionment to foreign affiliates for joint costs wherever there is a mutual benefit, based on the relative benefits. Apportionment of costs may be done by i cost­ sharing, based on the estimated share in the benefits of each affiliate; in the case of production­related services, this may be done by a mark­up on the price of goods sold; or ii `cost­funding ¶ by contributions from each affiliate based for example on gross turnover sometimes known as the `fixed­key ¶method. The difficulty is that these methods essentially entail allocation of costs by formula, which runs counter to the arms length approach. Nevertheless, following the 1979 OECD report the business lobby groups pressed for acceptance of cost contribution arrangements. In its more detailed report of 1984, the OECD Committee went some way towards accepting the arguments of TNCs for these indirect methods, although several countries still had significant reservations, particularly to cost­funding, where the charges may be unrelated to the benefit received. This report accepted that even 1 This is likely to mean the comparable rate in the country of the recipient rather than of the provider, since deductibility is decided by the source country. though cost­sharing does not correspond to arrangements applied between unrelated parties, it may nevertheless be justified by the `special situation of MNEs ¶OECD 1984­III, para.63. It laid down principles, described as non­exhaustive, which would facilitate the acceptance of a cost­sharing arrangement: it should be established in advance by a clearly formulated and binding contract, covering all affiliates which might be expected to benefit, observed consistently over several years and modified as soon as there is any relevant alteration in the activities of group members; the costs should be established by accepted accounting principles, and should entitle contributors to receive those services without any other payment OECD 1984­III, para 67. This goes some way towards accepting the proportionate allocation of joint costs within an international company group. However, many countries, especially those which have been mainly recipients of direct investment, are still very restrictive in their allowance of deduction for joint costs. The German rules permit cost­sharing contracts, but they are quite strict in requiring proof of actual services performed, although where the volume fluctuates over several years an average charge may be appropriate Rule 6.2.3. They also exclude a large category of costs coming within the `shareholder costs ¶of the parent company. This concept has proved controversial. The OECD Committee distinguished between a loose and decentralised firm where the role of the parent may be that of monitoring its investments, and an integrated and more centralised TNC. In the latter case, the managerial and coordination activities of the parent could be regarded as generating `extra profits ¶which accrue primarily to the subsidiaries and only indirectly to the parent, and therefore should be shared as a joint cost. Some OECD members however rejected this, and the issue was left to be dealt with bilaterally OECD 1984­III, paras 33­43. Finally, there is the question of whether a service charge can include a profit element. The OECD Committee concluded that a profit mark­up is always appropriate in some cases, particularly where the provider is specially established to perform such services, or is particularly capable of them and they are especially beneficial to the recipient. When a direct fee­for­service is charged based on market prices, the charge will presumably include a profit element, unless the provider is operating below capacity. However, when the charge is indirect and based on cost­sharing, it is argued that since there is no risk, a profit element is inappropriate. This is the position taken in the German rules, but they permit an allowance for the cost of capital invested. 3.e Intangibles and Synergy Profits The category of intangibles covers a very broad range of assets, related both to production and to marketing, and generally originating from knowledge, information or skills. They may be legally recognised as forms of property, such as patents, copyrights, designs, and trademarks; or they may be other types of legally protected rights, such as confidential data and business know­how, which are independent of the services of a particular individual and therefore belong to the enterprise. 1 Intangibles are especially important to TNCs, whose pre­eminence is often due to technological advances, product differentiation, market positioning or distinctive managerial methods. The transfer of rights to intangibles has two aspects: a payment for their value, and the acquisition of an asset which can generate an income stream. Therefore, such transfers are especially sensitive, since they can have a major effect on the distribution of the profits of a TNC. As with interest and central service charges, the source country will be concerned to ensure that royalties or fees payable for intangibles do not excessively reduce taxable business profits. Some countries, notably in Latin America, do not allow deductibility of royalties paid abroad, but this is only possible if the recipient is not a national of a treaty state, since the model treaties require non­discrimination. However, all countries frequently monitor carefully the extent and level of payments made for use of intellectual property rights, especially if made to low­tax countries: for example, Germany collects data on such payments especially to Switzerland and Liechtenstein, and maintains files for comparative purposes Germany, Bundestag 1986, 14­15. At the same time, the home country of a TNC will wish to ensure an adequate contribution or return from its foreign affiliates for the use of such assets. In addition, an intangible may be transferred to a subsidiary located in a country with a convenient tax regime: indeed, many countries offer specific incentives to try to attract high­technology investments. This can be hard for the home country of a TNC to counter through anti­base company or CFC measures: since the base­country affiliate will be carrying out a genuine manufacturing or service activity, it is hard to classify its profits as `passive income ¶ see Chapter 7 section 2 above. 2 Therefore, the question will often be whether the price paid for the intangible fairly reflects its value in relation to the profits it helps to generate. Although an intangible is often a specific identifiable property right, its particular nature as intellectual, scientific or artistic property gives it peculiar cost and profit characteristics. The essential problem is uncertainty and risk. It is necessary to make large up­front investments for basic scientific research and other creative activity such as design, or of even larger sums for applying these ideas to production or marketing, while no definite, measurable gains can be anticipated. Thus, At each stage in the R D process, the final commercial benefits to be derived remain uncertain ... and the degree of risk involved makes it difficult to estimate benefits from the outlays made which, in the event that some R D projects prove successful commercially, will only materialise in the future. Moreover, MNEs will 1 See the definition in the US tax code s.936h3b. 2 Indeed, a special problem has been posed for the US by US corporations transferring intangibles to subsidiaries in Puerto Rico, to take advantage of tax incentives enacted by the US itself for possessions corporations. understandably seek to recoup the cost of financing unsuccessful research from the results of successful research. OECD 1979 para.80. These characteristics create significant difficulties in establishing an arms length price for an intangible. First, it can be especially difficult to establish an arms length price based on a comparable, since the value of intangibles frequently lies in their difference or even uniqueness. Nevertheless, some authorities accept that the best that can be done is to evaluate the appropriateness of the terms of a transfer of intangibles in a general way, by comparison with similar transactions in the same industry, and perhaps using as a check the profit produced for the transferee over a period of time. Thus, the Italian Ministry of Finance in 1980 provided, for certainty and speed of administration, `safe harbour ¶ guide­lines for patent royalties, of up to 2 of sales provided there is a prior written contract, and evidence of actual benefit; over 2 and up to 5 could be acceptable for special factors originality, obsolescence, exclusivity etc; and over 5 of sales only in exceptional circumstances such as the high technological level of the industry Italy, Ministry of Finance 1980. This type of approach offers an easily administrable rule, but may prove inflexible. Where an adequate comparable cannot be found, it is difficult to fix an arms length price based on cost, since `the actual open market price of intangible property is not related in a consistent manner to the costs involved in developing it ¶OECD 1979 para. 100. The OECD report therefore concluded that cost can at best provide some guidance as to the lower limit of a price. Alternatively, some TNCs have used cost­contribution arrangements, under which affiliates pay a share of the direct and indirect costs of a research programme based on the proportionate benefit they are expected to derive, in exchange for rights in the intangibles that are produced. These arrangements are similar to the cost­sharing or cost­funding methods used for central services discussed above; and they raise, even more acutely, the same questions: is there a clear benefit for the recipient, and does the method used to fix the contribution fairly reflect the expected benefit. The 1979 OECD report pointed out that `such arrangements might open up opportunities for profit transfers disguised as deductions for costs ¶ and stressed the `need for a strict interpretation of the notion of real benefit ¶ to ensure that participants only pay for expenditure which is really in their interests OECD 1979 para.115. The German rules allow cost­contribution contracts for research, on the same conditions as for central administrative services: they should be based on a clear prior contract, distributing actual identifiable direct and indirect costs by a recognised accounting method, in exchange for rights to the intangibles produced without any supplementary payments, and making an apportionment of costs based on the benefits anticipated Section 7 of the 1983 rules. The US regulations of 1968 also allowed cost­sharing agreements, provided they are in writing, and reflect a good faith effort by participants to bear their proportion of costs and risks on an arms length basis regulation 1­482­2d4. 1 The 1988 White Paper conceded that cost­sharing could still be acceptable after the enactment of the `commensurate with income ¶ requirement by Congress see below, but only under strict conditions. In particular, the sharing of costs should not be confined to single products which might enable a foreign subsidiary to benefit from a high­profit item without paying for lower­profit or unsuccessful research, but should normally be based on product areas defined at least by 3­digit SIC Standard Industrial Classification code; although either tax­ payer or IRS could show that a narrower or broader agreement is more appropriate US Treasury 1988 chs. 12 and 13. Thus, the problems involved in establishing arms length prices for specific intra­firm transactions are at their most intractable when it comes to intangibles. Intangibles are likely to be distinct if not unique, throwing into doubt the validity of comparisons with similar items, if any are available on the market. Pricing based on costs is extremely difficult, since they are essentially joint cost factors with uncertain outcomes, so that either fixing a price based on cost, or allocating cost contributions in relation to anticipated benefit, is likely to be arbitrary. Above all, intangibles most clearly raise the problem of the allocation of profits from synergy: the additional profits generated for an integrated firm by reason of its features as an organization, and not attributable specifically to any of its parts. Like joint costs, synergy profits cut to the heart of the separate enterprise approach to the taxation of international company groups, since such profits by definition cannot be attributed to one particular affiliate. Both problems also entail the politicization of the transfer pricing question, since the issue posed is the allocation of profit rather than its attribution. If a country accepts deductibility of payments from a local operating subsidiary of a TNC for intangibles owned or developed elsewhere, it will be allowing a deduction from gross profits, or essentially a tax subsidy, for research based abroad. Furthermore, countries compete in offering tax advantages for high­ technology operations, and a TNC may transfer rights to intangibles to a subsidiary located in a country with an advantageous tax regime, so that the TNC benefits from lower taxation of the additional monopoly profit. Thus for example, the German authorities have been particularly vigilant in scrutinising technology royalties and licence fees paid to foreign related companies, especially to those located in Switzerland and Liechtenstein German Bundestag 1986, p.14. 1 These criteria were expressed in much more general terms than the draft regulations proposed in 1966, which had put forward detailed rules: 31 Fed.Reg. 10394. 3.f The Commensurate with Income Standard and the US White Paper The experience of these difficulties led to proposals by the US authorities in 1988 for a new approach. As has already been mentioned, there is a long history of political concern about the effectiveness of US taxation of US­based TNCs. Specifically, when Congress attacked tax deferral with the Subpart F provisions in 1962, it considered proposals to require formula apportionment if a TNC could not prove comparable arms length transfer prices; these were withdrawn following administration assurances that the desired result could be obtained by increased enforcement of s.482. Nevertheless, there has been continuing debate about the effectiveness of enforcement of the section and of the 1968 regulations, which reached a peak with the GAO report of 1981. Although the IRS responded vigorously to those criticisms, both the IRS and the Treasury were aware of underlying problems with the arms length method. These concerns resulted in a provision in the mammoth Tax Reform Act of 1986 amending s.482, the first substantial amendment of that section since 1928. This provision, popularly referred to as the `superroyalty ¶ added the following final sentence to the section: In the case of any transfer or license of intangible property within the meaning of s.936h3B, the income with respect to such transfer or license shall be commensurate with the income attributable to the intangible. At the same time, Congress referred to continuing unresolved difficulties with s.482, and asked for a comprehensive study of intercompany pricing rules giving careful consideration to whether the existing regulations should be amended. The result was a bulky discussion document issued by the Treasury and IRS in October 1988, `A Study of Intercompany Pricing ¶ This study identified two major problems in the enforcement of s.482. First was the difficulty of obtaining adequate information to evaluate pricing, leading to long delays in closing cases. It argued that the burden was on taxpayers to document established pricing policies, but evidence from IRS examiners showed that most firms sought to justify the prices charged only when challenged, and by reference to whatever method most closely approximated to those prices. 1 The main problem, to which the bulk of the study was addressed, was the pricing of intangibles. Although this, and especially the problem of `high profit intangibles ¶ was the explicit focus, 1 The study urged a more aggressive use of administrative summons procedures by examiners see Chapter 10.2.a below. It also pointed to the need for special scrutiny of the growing volume of direct investment into the US, in view of evidence of substantially lower profit levels being reported for foreign­owned TNCs US Treasury 1988 p.15. Although this point was given less attention initially than the remaining major parts of the report, the issue of the apparently low tax liability of foreign investors in the US later came to the fore, as mentioned in section 2.b above. the study had far­reaching implications, since the pricing of intangibles relates closely to charging for central services generally, and is also often `bundled ¶in the price of tangibles. Hence, the question of intangibles is central to the structure of intra­firm relationships. The White Paper articulated the implications of the `commensurate with income ¶ requirement inserted by Congress into s.482, while arguing that it is compatible with the arms length standard. To do so, it put forward a new pricing method, based on rate of return, called the Basic Arms Length Return Method quickly labelled BALRM or the `ballroom ¶ method, and severely circumscribed the application of other methods. i Restriction of the CUP Method As to comparables, the study argued that the CUP method can only be relied upon if an `exact comparable ¶ can be found. An exact comparable would be a transaction transferring the same intangible to an unrelated party. Since such transfers are usually exclusive, it need not be in respect of the same market. However, there must be substantial comparability both of external factors overall size of the market, degree of competition, collateral or continuing relationships, and of internal factors the terms of the contract. An `inexact comparable ¶ involving a similar intangible, may be useful to provide comparative information, but cannot be the sole basis to determine pricing. It can only be relied upon if `the differences between it and the related party transaction can be reflected by a reasonable number of adjustments that have definite and ascertainable effects on the terms of the arrangement ¶US Treasury 1988, p.91. The study was especially critical of reliance on industry standards or averages. This would severely restrict the use of the CUP method. The White Paper stated that comparability would be easy to demonstrate in cases of widely available technology such as pocket calculators, digital watches or microwave ovens. But such technology is unlikely to be transferred within an integrated TNC unless it has distinct properties: for example, ball­point pen technology is widely available, but a company such as Parker Pen still commands premium prices based on unique intangibles. Thus, in a case following the White Paper but decided under the pre­1986 law, the Tax Court took the view that non­exclusive licences to soft contact lens manufacturing technology, granted by Bausch Lomb to its production subsidiary in Ireland, could be evaluated on the basis of comparables, and rejected the IRS view that the unrelated party prices were not fully comparable because of the lower volumes involved Bausch Lomb v. Commr. 1989. 1 These examples also demonstrate that the `exact comparability ¶requirement applies to sales of tangibles which embody an intangible. If components, ingredients or subassemblies embody design or production technology, or marketing intangibles 1 This decision was upheld by the 2nd Circuit Court of Appeals on 14 May 1991. such as brand identification, the requirement that a comparable be `exact ¶will reduce the application of the CUP method for tangibles also. ii Periodic Adjustments The commensurate with income requirement also entails that the return for a transfer of an intangible must normally be subject to periodic adjustment, to reflect the actual profit earned from it. The study defends this as compatible with arms length, citing evidence that unrelated parties also renegotiate terms, either on the basis of explicit clauses providing for renegotiation or by using termination clauses to do so. 1 Hence, payments fixed by reference to either exact or inexact comparables must be reviewed, and adjusted if there has been a significant unforeseen change in profitability due to subsequent events. However, periodic adjustment can be avoided if the taxpayer can point to an exactly comparable arrangement which does not permit adjustment US Treasury 1988, p.64. iii The Arms Length Return Method The study put great reliance on the proposed new `arms length return ¶method, which should be used if there is no exact comparable and to adjust a price based on an inexact comparable. This identifies the assets and factors of production employed by each related party, and allocates to them a `market ¶rate of return. The first step is a functional analysis, breaking down the activities carried out by the parties into their component functions; next, those functions with measurable assets or production factors can be assigned a market rate of return on the value of those assets. The residual profit is then normally assigned to the parent company, since it is considered to be attributable to the non­measurable intangible assets. The reasoning is that many of the functions carried out by affiliated companies are also carried out by unrelated firms, and information will be available about their normal rates of return on measurable factors plant, labour, equipment, working capital and routine know­ how. In that sense, this is an arms length method, since it is attributing profits based on those of comparable firms in the industry. It therefore resembles the profit comparison approaches based on `empirical ¶ methods or the use of `coefficients¶ long permitted in international practice for permanent establishments, and also practised by several countries in relation to subsidiaries, as a fall­back if evidence based on market transactions is not available as discussed earlier in this chapter. 1 SSHQGL[RIWKH:KLWH3DSHUSURYLGHVDCYHU\SUHOLPLQDU\¶DQDO\VLVRIXQUHODWHGSDUW\OLFHQFH agreements obtained from the files of the Securities and Exchange Commission, and a literature review, to show that licensors seek to secure a risk­free return commensurate with actual profitability. But the study concedes that the Congressional enactment of the commensurate with income requirement was a legislative rejection of the tax court decision in French v. Commr. 1973, which held that a long­term fixed­rate royalty agreement could not be adjusted under s.482 to take into account subsequent unforeseen events. The major difficulty with the BALRM, however, is the way it separates `measurable ¶ factors, to which a normal rate of return is assigned, and then assumes that the residual profit is attributable to the parent company. This was immediately criticised by some tax specialists as an attempt by the IRS to treat foreign subsidiaries as `contract manufacturers ¶ despite tax court decisions rejecting such a result Fuller 1988. It certainly seems that the rate of return method proposed in the white paper has grown out of the analyses by IRS economists of company accounts in s.482 cases, using a `profit split ¶first to support a price adjustment and finally, in the Eli Lilly case, as the basis for the adjustment. In Eli Lilly, the Circuit Court upheld the Tax Courts `profit split ¶ methodology, which `divides combined revenues based on an ad hoc assessment of the contributions of the assets and activities of the commonly controlled enterprises ¶Eli Lilly v. Commr. 1988, p.871. However, in that case, on the view taken by the courts both the related parties owned some intangibles. Lilly had set up a manufacturing subsidiary in Puerto Rico, to which it had transferred valuable drug patents in exchange for stock, while retaining the US marketing intangibles trade names etc.. The IRS challenged the transfer, on the ground that if the parties had been unrelated the transferor would have required payments for the continuing research and development programme; this would have meant treating the Puerto Rico subsidiary as a contract manufacturer. The courts, however, accepted that the stock transfer was valuable consideration, and judged that Eli Lillys high profits from the US sales of the drug, based on its US marketing intangibles, provided sufficient funding of the research effort. But the Tax Court first re­evaluated the costs attributable to each party and allocated an appropriate rate of return on those costs including a return to the US parent of 100 on marketing costs; and then adjusted the profit split on the residual profit to a ratio of 45­55 Lilly had allowed 40­60, reflecting its view of the relative contributions of Lilly USs marketing intangibles and Lilly PRs production intangibles. 1 Hence, although the US courts have not yet accepted a profit split which treats a foreign subsidiary as a mere contract manufacturer, this is because they have accepted that the subsidiary validly owned some intangibles see also Searle v. Commr. 1987. The White Paper argued that the profit splits accepted by the courts in previous cases were arbitrary or had `no discernible rationale ¶ and this was the reason for developing a `careful functional analysis ¶US Treasury 1988, p.39, i.e. the BALRM 1 The Circuit Court upheld all the Tax Courts adjustments except a charge to Lilly P.R. for research and development expenses, which it considered incompatible with the acceptance of the patents­for­ stock transfer: ibid. p.871. In response to the Lilly case s.936h was added to the Tax Code in 1982, so that possessions corporations must split the revenue from intangibles at least 50­50 with their US parent. The 1986 Tax Reform Act, in adding the commensurate with income requirement to s.482, also amended s.367d2 so that recognition of a transfer of intangibles is conditional on payments CFRPPHQVXUDWHZLWKWKHLQFRPHDWWULEXWDEOHWRWKHLQWDQJLEOH¶ method. The difficulty is that functional analysis is only precise in relation to `measurable ¶factors. If there are synergy profits, functional analysis cannot provide a precise allocation. This the White Paper accepts, in its discussion of what it calls the `profit split addition to the basic arms length return method ¶ It countenances that, where a large TNC has foreign subsidiaries that perform complex functions, take significant risks and own significant assets, it would be appropriate to divide the `residual ¶or synergy profits according to the relative value of the intangibles contributed or risks assumed by the parties. It is conceded that `it is easier to state this principle than to describe in detail how it is to be applied in practice ¶ and that `splitting the intangible income in such cases will be largely a matter of judgment ¶US Treasury 1988, p.101. This is perhaps a concession to the USs main treaty partners, some of which may be reluctant to accept an approach which allocates the entirety of the synergy profits of US TNCs to the parent company. 1 Thus, the full rigour of the BALRM may be confined to subsidiaries with high­profit intangibles in convenient low­tax locations, such as Ireland or Singapore, or operating subsidiaries in relatively small markets. In principle, the application of the BALRM should also limit US taxation of foreign­owned TNCs Carlson, Fogarasi Gordon 1988, and this is becoming more significant as the growth of direct investment into the USA continues to even out the previous imbalance. However, this would depend on the willingness of the US authorities to accept that foreign­owned subsidiaries in the US do not own significant intangibles. It is more likely that due to the size and complexity of the US market, foreign­owned companies would be regarded as substantially distinct operations from their parents, thus justifying an allocation to them of an appropriate proportion of the residual profits. Equally, the other main OECD countries may also argue that there is a difference between major TNC affiliates with substantial operations and their own independent know­how, and smaller subsidiaries operating in restricted markets, or carrying out specific functions of a `contract manufacturer ¶kind. The rate of return method based on functional analysis represents a significant retreat from the attempt to adjust prices of individual transactions mainly by finding comparables. It may be that the identification of functions will provide a more precise and flexible basis for evaluating profit than other `empirical ¶methods that have been used, such as gross margin on sales. It also makes it explicit that, at least in the case of integrated, research­based TNCs, there is likely to be a significant element of `residual ¶ profit attributable to the organization as such, which must either be assigned to the parent, or allocated among the main affiliates according to some measure of their contributions to the innovations and risks of the organization. But it 1 The official foreign government responses to the White Paper were confidential and not published, although many of the relevant officials have expressed their personal views in symposia: see e.g. De Hosson ed 1989. does not resolve many of the problems which historically made the tax authorities fight shy of explicitly adopting a profit­split method. Hence, it is not much help in evaluating the profit split between the related parties: by focussing on `measurable ¶ functions, it merely assumes that synergy profits accrue to the ultimate parent. Due to their broader political and economic implications, there is likely to be continued reluctance to tackle these questions explicitly and publicly, and a continuing insistence that the normal and primary test of transfer prices is that of independent parties dealing at arms length. Indeed, the White Paper was fiercely criticised by the Taxation Commission of the International Chamber of Commerce, as entailing `fundamental deviations from the arms length principle ¶ Professor Wolfgang Ritter, responding to the White Paper on behalf of that Commission, wrote that the Basic Arms Length Return Method: is nothing but another form of unitary taxation, now recognised by the US Federal Authorities, as a result of the international outcry which it provoked, to be unfair. BALRM analysis does exactly what unitary taxation does: instead of looking at the actual prices charged between related entities, it looks at the total profits earned by a number of entities and divides them up by reference to criteria determined by academic economists with no experience of the real business world. Ritter, in De Hosson 1989, p.73.

4. Transfer Prices in Theory and Practice